Dynamic Education Signaling with Dropout, Second Version

We study students' dropout behavior and its consequences in a dynamic signaling model. Workers pay an education cost per unit of time and cannot commit to a fixed education length. Workers face an exogenous dropout risk before graduation. Since low-productivity students' cost is high, pooling with early dropouts helps them to avoid a high education cost. In equilibrium, low-productivity students choose to endogenously drop out over time, so the productivity of students in college increases along the education process. We find that the maximum education length is decreasing in the prior about a student being highly productive. We characterize the joint dynamics of returns to education and the dropout rate and provide an explanation of the declining dropout rate over the time students spend in school. We also extend the baseline model by allowing human capital accumulation and show that the dynamics of the dropout rate are helpful in decomposing the returns to education into the signaling effect and the human capital accumulation effect.